This Article Abstract Résumé de cet Article Full Text (PDF) Submit a scholarly reply Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Huang, H. Articles by Liu, J. Search for Related Content PubMed PubMed Citation Articles by Huang, H. Articles by Liu, J.

Effect of ischemic post-conditioning on spinal cord ischemic-reperfusion injury in rabbits

[Effet du post-conditionnement ischémique sur les lésions d’ischémie-reperfusion de la moelle épinière chez le lapin]

From the Department of Anesthesiology, West China Hospital, Sichuan University, State Key Laboratory of Biotherapy of Cancer, Chengdu, Sichuan, China.

Address correspondence to: Dr. Jin Liu, Professor and Chairman, Department of Anesthesiology, West China Hospital, Sichuan University, Chengdu, Sichuan 610041, P.R. China. Phone: 86-28-8542-2520; Fax: 86-28-8542-3591; E-mail: huanghan1981{at}hotmail.com

	Abstract

TOP Abstract Introduction Material and methods Results Discussion References

 
Objectives: To investigate the potential protective effect of ischemic post-conditioning (Post-con) on ischemia-reperfusion injury of the rabbit spinal cord, and to determine if there is an additive neuroprotective effect when ischemic preconditioning (IPC) and Post-con are combined.

Methods: Forty New Zealand white rabbits were randomly divided into four groups: group Control (C; n = 10), aortic occlusion (AOC; for 30 min; group IPC (n = 10) three cycles of three-minute AOC plus three-minute reperfusion before the 30-min AOC; group Post-con (n = 10), three cycles of three-minute reperfusion plus three-minute AOC immediately upon reperfusion after 30-min AOC; group IPC+Post-con (n = 10), where animals were subjected to both IPC and Post-con. At six hours, 24 hr and 48 hr following reperfusion, neurological function was assessed according to Tarlov scores, and at 48 hr, the spinal cords were procured for the histopathologic evaluation, by comparing the number of intact -motor neurons in the anterior horn.

Results: The median count (and quartiles) of intact -motor neurons was greatest in group Post-con 73 (69–76) and group IPC+Post-con 29 (22–42) compared to the numbers of viable -motor neurons in groups C 6 (4–9) and IPC 15 (11–18) (P < 0.001). The numbers of animals who developed paraplegia according to Tarlov criteria were 7/10 in groups Post-con and IPC+Post-con, compared to 9/10 animals in each of groups C and IPC.

Conclusions: This laboratory investigation provides histological evidence that Post-con may protect the spinal cord from moderate to severe ischemia reperfusion injury. Ischemic preconditioning conferred no additional benefits in this rabbit model. The results have potential clinical implications for patients undergoing thoracoabdominal aortic reconstructive surgery.

	Introduction

TOP Abstract Introduction Material and methods Results Discussion References

 
PARAPLEGIA following descending thoracic or thoracoabdominal aortic reconstructive surgery remains one of the most devastating complications of this high-risk operation. Although a number of strategies have been used to reduce the risk of spinal cord injury associated with this surgery, including induced hypothermia and cerebrospinal fluid drainage, the therapeutic benefits of these interventions remain uncertain.

Since ischemic preconditioning (IPC) was first reported by Murry et al.¹ to delay lethal cell injury in ischemic myocardium, this powerful endogenous neuroprotective mechanism has also been identified in other organs, including the spinal cord.² However, the efficacy of IPC for spinal cord protection remains uncertain,³ a fact which has limited its widespread clinical application.

There are established mechanisms underlying IPC which warrant further evaluation. For example, IPC could reduce the production of oxygen free radicals⁴ that are toxic to the tissues, increase the expression of heat shock proteins that inhibits protein degradation, ⁵ and promotes the opening of the K_ATP channel⁶ to decrease the inward flux of Ca²⁺. Some of these mechanisms have also been found to exist in ischemic post-conditioning (Post-con), which may also be protective of ischemic myocardium,⁷ indicating that the reperfusion itself could initiate endogenous cellular protective mechanisms. However, just as its name implies, IPC must be initiated before the ischemic insult. Unfortunately, in most cases, ischemic events are unpredictable. In contrast, Post-con has the potential to be introduced immediately following an ischemic event at the onset of reperfusion, which could result in a potentially beneficial therapeutic benefit. Accordingly, Post-con could have a much broader clinic application compared to IPC.

The present study was designed to assess the effects of Post-con on neurological and histopathological outcomes in a rabbit model of severe spinal cord ischemia, in comparison to the classical IPC. In addition, we also sought to determine a possible additive effect when the two neuroprotective strategies are combined.

	Material and methods

TOP Abstract Introduction Material and methods Results Discussion References

 
Animal care
All experimental procedures and protocols used in this investigation were reviewed and approved by the Animal Research Committee of Sichuan University. Forty New Zealand white rabbits were studied, each animal weighing between 2.0 and 2.5 kg. The animals were housed with free access to food and water. All rabbits were neurologically intact before anesthesia and the surgical procedure.

Surgical preparation
The rabbits were fasted overnight before surgery and were anesthetized with ketamine 25 mg·kg^–1 im and midazolam 1 mg im. A 22-G iv catheter was then placed in an ear vein for iv administration of fluids and drugs. After endotracheal intubation, vecuronium 1 mg iv was administered and the rabbits were placed on a ventilator (inspired oxygen fraction, 100%) with an initial tidal volume of 40 mL at a rate of 40 cycles·min^–1. Ventilatory parameters were subsequently adjusted to maintain the PaO₂, PaCO₂ and pH values within the normal physiologic range, as confirmed by the arterial blood gas analysis. An ear artery was cannulated with a 22-G catheter to monitor arterial blood pressure (Spacelabs Medical, Inc., Redmond, WA, USA). Heart rate was calculated from the blood pressure wave by the pressure monitor. Hemodynamic parameters were monitored throughout the experiment and were recorded after surgical preparation (baseline), at the mid-point of IPC if any, at the mid-point of the 30-min period of ischemia, at the mid-point of Post-con if any, and 30 min after restored reperfusion, respectively. Each animal’s tail was prepared for an arterial pulse oximeter probe, and a rectal thermometer was inserted as a guide to maintaining body temperature around 38°C throughout the operation, with the aid of a heating lamp.

Following a midline laparotomy, the abdominal aorta was carefully exposed. Five minutes before occlusion, heparin 1 mg·kg^–1 iv was administered. A bulldog clamp was then placed across the aorta, just below the left renal artery, to produce abdominal aortic occlusion (AOC). The clamp remained in place for 30 min while the response was being monitored. Throughout the experiment, lactated Ringer’s solution was infused at 10 mL·kg^–1·hr^–1 and additional doses of midazolam, vecuronium and fentanyl were administered as required to maintain the desired level of anesthesia. After measurements were completed, the aorta was declamped, the laparotomy incision was closed, and animals were weaned from the ventilator and allowed to recover from anesthesia before being returned to their cages for postoperative care.

Experimental protocol
Animals were randomly assigned to one of four groups (Figure 1) according to a computer-generated random number schedule: 1) control (C, n = 10): the aorta was occluded for 30 min; 2) ischemic preconditioning (IPC, n = 10): before the 30 min AOC, the aorta was occluded for three minutes followed by three minutes of reperfusion, repeated for three cycles; 3) ischemic post-conditioning (Post-con, n = 10): after the 30 min AOC, reperfusion was initiated for three minutes followed by three minutes of re-occlusion, repeated for three cycles; 4) ischemic preconditioning plus post-conditioning (IPC+Post-con, n = 10): the above IPC and Post-con protocols were combined.

View larger version (15K): [in this window] [in a new window]   FIGURE 1 Experimental protocol: [control (C, n = 10)]; the aorta was reversibly occluded for 30 min; ischemic preconditioning (IPC, n = 10): before the 30 min period of aortic occlusion (AOC), the aorta was occluded for three minutes followed by three minutes of reperfusion, repeated for three cycles; ischemic post-conditioning (Post-con, n = 10): after the 30 min AOC, reperfusion was initiated for three minutes followed by three minutes of reocclusion, repeated for three cycles; Ischemic preconditioning plus post-conditioning (IPC+Post-con, n = 10): the above IPC and Post-con protocols were combined.

Neurological evaluation
At six hours, 24 hr and 48 hr following reperfusion, the hind limb motor function was assessed by a trained observer. Assessment was done according to the modified Tarlov criteria:⁸ 0 = paraplegia with no lower-extremity movement; 1 = poor lower-extremity movement but unresistant to gravity; 2 = good lower-extremity motor function with resistance to gravity, but unable to draw legs or hop; 3 = ability to draw leg and hop but not normally; 4 = normal lower-extremity motor function. Animals with Tarlov grades 0–2 were rated as being paraplegic, while grades 3–4 were considered non-paraplegic.

Histopathologic studies
Forty-eight hours after reperfusion, all animals were sacrificed and their spinal cord were immediately harvested and immersed in formalin for later histopathologic examination. Each segment of the spinal cord below L-5 (including L-5) was dissected in a 5-mm block. The blocks of tissues were embedded in paraffin. Transverse sections of each block were sliced for hematoxylin and eosin staining. One section from each block was evaluated by an experienced pathologist, who was unaware of the experimental conditions. Viable -motor neurons in the ventral horns were counted, and the results from all the sections of each spinal cord were aggregated to give a final result for each animal. A viable -motor neuron was defined according to the following criteria: fine granular cytoplasm with Nissl body and a soma diameter of 30–60 µm. Neurons with eosinophilic cytoplasm were considered to be either dead or necrotic (red neurons).

Statistical analysis
Hemodynamic data were compared using repeated measures analysis of variance (ANOVA) to test for group-time interactions, and when significant differences were identified, the Student-Neumann-Keuls test was applied post-hoc. Tarlov scores were compared via the Kruskal-Wallis test, followed by the Mann-Whitney U test where indicated. The incidence of paraplegia was compared by the Mantel Haensel test. Hemodynamic data are presented as means ± SD, Tarlov scores and the numbers of viable neuron are expressed as medians (and 25% and 75% quartiles), and the incidence of paraplegia is presented in absolute numbers. Data were analyzed using SPSS software (version 11.5, SPSS Inc, Chicago, IL, USA), and a P value < 0.05 was considered statistically significant.

	Results

TOP Abstract Introduction Material and methods Results Discussion References

 
Hemodynamics
Mean arterial pressure and heart rate were similar amongst the four groups at corresponding time periods (Table I). Throughout the duration of AOC, the pulsatile wave of S_PO₂ disappeared in all animals, confirming completeness of the AOC. Soon after aortic cross clamp release, S_PO₂ values returned to 100% with corresponding reappearance of normal oximetric waveforms, confirming reperfusion (data not shown). No differences in body temperature were observed, and anesthetic doses and volumes of infused fluid were similar in the four groups.

	Discussion

TOP Abstract Introduction Material and methods Results Discussion References

 
This study provides histopathological evidence that Post-con may protect the spinal cord against ischemia reperfusion injury, providing further evidence in support of a potential benefit of post-ischemic conditioning. ⁹ The underlying mechanisms by which Post-con may protect the spinal cord are unknown. What has been learned from cardiac studies is that post-conditioning can reduce the production of reactive oxygen species (ROS) and the inflammation caused by ROS.

The duration of post-conditioning ischemia is critical. Kin et al.¹⁰ demonstrated that the protective effect of Post-con would be lost if the first reperfusion event lasted longer than one minute, because of a significant burst of oxygen-derived free radicals generated within the first few minutes of reperfusion. However, compared to Kin’s study (in a rat model examining the myocardial response to three cycles of one-minute reperfusion and one minute of ischemia), the reperfusion in our study was longer (involving the spinal cord in rabbits, with three cycles of three-minute reperfusion and three minutes of ischemia), although histological evidence of a protective effect still existed. Furthermore, in Jiang et al.’s study,⁹ the neuroprotective effect persisted, even though the first episode of reperfusion lasted as long as five minutes.

There are several possible explanations to account for the variances between these studies. First, xanthine oxidase in rats¹¹ is considerably more active than in the rabbit, resulting in greater free oxygen radical production, and potentially greater exposure to tissue injury in rats compared to rabbits. In addition, the heart is metabolically more active than the spinal cord. As more oxygen free radicals are generated in tissues of higher metabolic activity, the myocardium may be more susceptible to injury by the toxic free radicals. Lips et al.¹² have also reported that blocking of the N-methyl-D-aspartate (NMDA) receptor may protect the ischemic spinal cord against ischemia reperfusion injury, although the potential role of the NMDA receptor in Post-con has not been evaluated. Considering the differences between species and tissues, the mechanism of the potential Post-con spinal cord neuroprotection observed in rabbits warrants further investigation.

In group Post-con, histological evidence of neuroprotection did not translate into a significant reduction in the incidence of paraplegia according to Tarlov scores. One possible reason is that the injury severity induced by the protocol in this rabbit model exceeded a duration such that Post-con could only improve the injury to a degree that was just between the threshold of histopathologic change and functional change. It has been demonstrated previously that cellular function is compromised well before the morphologic changes of cell death.¹³ Compared to Sakurai et al.’s¹⁴ study in which the 15-min ischemia produced only mild injury in the C group (there were no animals with Tarlov scores of 0 or 1 in this group), our 30- min period of ischemia caused more severe injury (in the C group, nine of ten animals had paraplegia scores of 0 according to Tarlov’s criteria, data not shown). We chose this duration of ischemia to provide greater potential clinical relevance. As pointed out by Svensson et al.¹⁵ mild paraparesis is likely to resolve and many patients can regain their preoperative motor function. Ueno et al.¹⁶ also reported that the correlation between histological findings and functional outcomes are variable. Some authors¹⁷ report that the protective effect of IPC could be transient, disappearing within seven days following reperfusion. However, Jiang et.al.⁹ have demonstrated that the histopathogic status in their Post-con group was better than that observed in the C group after a ten-day period of reperfusion.

There have been very few reports to date examining the effects of IPC on the spinal cord. Zvara et al.¹⁸ demonstrated that IPC induced by three minutes of ischemia reduced neurological injury resulting from a subsequent 30-min period of ischemia. Sirin et al.¹⁹ also investigated the effects of IPC with five minutes of ischemia on spinal cord injury mediated by a 20-min ischemic insult, and demonstrated that IPC reduces the extent of spinal cord injury. In the present study, a lack of response to IPC with three cycles may explain the lack of protection by IPC, due to the frequency of preconditioning. Iliodromitis et al.²⁰ showed that six to eight cycles of brief ischemia result in a loss of efficacy of IPC in rabbit myocardium, but four cycles were still protective in their study. Ischemic preconditioning and Post-con have some overlapping elements in their signal pathways, which has been confirmed in the myocardium.²¹ The effect of Post-con was also dampened in group IPC+ Post-con, suggesting there may be an interaction between the two ischemic conditionings. However, it remains unclear as to why an increasing number of preconditioning cycles may dampen the protective phenomenon. Interestingly, the preconditioning itself would induce no harm to the tissues, since the results in the IPC group were not significantly different from the results in group C, an observation similar to that observed by Iliodromitis et al.²⁰

Another possible explanation for the absence of neuroprotection in the IPC group of our study is that the reperfusion interval between IPC and the subsequent ischemic event was only three minutes, much shorter than that of the other "protective" IPC protocols. ^18,22 Recently, Contreras et al.²³ demonstrated the importance of the reperfusion interval by monitoring the somatosensory evoked potentials. Further experiments using different cycles and durations of IPC and reperfusion interval are required to address the observed between-study variances.

In our study, transient hypotension was observed in all animals after releasing the aortic clamp, however, mean arterial pressure returned spontaneously to baseline values within one minute without requiring further intervention. Other investigators have reported similar hemodynamic responses.^12,24 In contrast to Toumpoulis et al.’s study,² hypotension should not have been a confounding variable in the interpretation of our data.

There are several methodological limitations of this study which merit comment. Given the fact that the burst of toxic free radicals peaks at four to seven minutes during reperfusion in some models,²⁵ and most reports evaluating post-conditioning responses adopt three cycles, we used a protocol of three cycles of three-minute reperfusion following three-minute ischemia in group Post-con. For the convenience of symmetric comparison between the two conditionings, we also used three cycles of three-minute ischemia for preconditioning, which may not be an optimal protocol. Further studies are needed to determine the optimum number of cycles and cycle duration for both IPC and Post-con. Further, while the study design was able to control a number of variables, as a preliminary laboratory investigation of spinal cord histological responses to post-ischemic conditioning it was not powered to detect differences in functional outcomes between groups. While the numbers of paraplegic animals were numerically less in groups Post-con and IPC+Post-con, the study was not powered to detect between-groups differences in Tarlov scores. The data from this study will be useful to design larger experiments to specifically address the duration and frequency of post-conditioning cycles and the effects on functional outcomes.

In conclusion, this laboratory investigation provides histological evidence that Post-con may protect the spinal cord from moderate to severe ischemia reperfusion injury. Ischemic preconditioning conferred no additional benefits in this rabbit model. The results have potential clinical implications for patients undergoing thoracoabdominal aortic reconstructive surgery, suggesting that further studies are well warranted.

View larger version (171K): [in this window] [in a new window]   FIGURE 2 Representative photomicrographs of spinal cord sections stained with hematoxylin and eosin. A number of viable -motor neuron cells (arrow) were preser ved in the ischemic preconditioning plus post-conditioning group (IPC+Post-con), whereas prominent vacuolization and eosinophilic red neurons (arrow head) were obser ved in the Control group (C) and group IPC (original magnification, x 200).

	Acknowledgments

	Footnotes

 
Presented in part at the 2005 Annual Meeting of the American Society of Anesthesiologists.

This study was partially funded by an operating grant from the 973 Program (2005CB522601), Beijing, P.R. China.

Accepted for publication August 18, 2006. Revision accepted October 19, 2006.

	References

TOP Abstract Introduction Material and methods Results Discussion References

 
1 Murry CE, Jennings RB, Reimer KA. Preconditioning with ischemia: a delay of lethal cell injury in ischemic myocardium. Circulation 1986; 74: 1124–36.

2 Toumpoulis IK, Anagnostopoulos CE, Drossos GE, Malamou-Mitsi VD, Pappa LS, Katritsis DG. Early ischemic preconditioning without hypotension prevents spinal cord injury caused by descending thoracic aortic occlusion. J Thorac Cardiovasc Surg 2003; 125: 1030–6. Comment in: Eur J Cardiothorac Surg 2003 24: 851–2; author reply 852–3.[Abstract/Free Full Text]

3 Ueno T, Chao ZL, Okazaki Y, Itoh T. The impact of ischaemic preconditioning on spinal cord blood flow and paraplegia. Cardiovasc Surg 2001; 9: 575–9.[Medline]

4 Vanden Hoek T, Becker LB, Shao ZH, Li CQ, Schumacker PT. Preconditioning in cardiomyocytes protects by attenuating oxidant stress at reperfusion. Circ Res 2000; 86: 541–8.[Abstract/Free Full Text]

5 Yellon DM, Latchman DS. Stress proteins and myocardial protection. J Mol Cell Cardiol 1992; 24: 113–24.[Medline]

6 D’hahan N, Moreau C, Prost AL, et al. Pharmacological plasticity of cardiac ATP-sensitive potassium channels toward diazoxide revealed by ADP. Proc Natl Acad Sci USA 1999; 96: 12162–7.[Abstract/Free Full Text]

7 Zhao ZQ, Corvera JS, Halkos ME, et al. Inhibition of myocardial injury by ischemic postconditioning during reperfusion: comparison with ischemic preconditioning. Am J Physiol Heart Circ Physiol 2003; 285: H579–88.[Abstract/Free Full Text]

8 Tarlov IM. Acute spinal cord compression paralysis. J Neurosurg 1972; 36: 10–20.[Medline]

9 Jiang X, Shi E, Nakajima Y, Sato S. Postconditioning, a series of brief interruptions of early reperfusion, prevents neurologic injury after spinal cord ischemia. Ann Surg 2006; 244: 148–53.[Medline]

10 Kin H, Zhao ZQ, Sun HY, et al. Postconditioning attenuates myocardial ischemia-reperfusion injury by inhibiting events in the early minutes of reperfusion. Cardiovasc Res 2004; 62: 74–85. Comment in: Cardiovasc Res 2004; 62: 4–6.[Abstract/Free Full Text]

11 Downey JM, Miura T, Eddy LJ, et al. Xanthine oxidase is not a source of free radicals in the ischemic rabbit heart. J Mol Cell Cardiol 1987; 19: 1053–60.[Medline]

12 Lips J, de Haan P, Bodewits P, et al. Neuroprotective effects of riluzole and ketamine during transient spinal cord ischemia in the rabbit. Anesthesiology 2000; 93: 1303–11.[Medline]

13 Richard N. Mitchell, Ramzi S. Cotran. Cell injury, adaptation, and death. In: Kumar V, Cotran RS, Robbins SL (Eds). Robbins Basic Pathology, 7th ed. Singapore: Elsevier; 2003: 3–31.

14 Sakurai M, Hayashi T, Abe K, Aoki M, Sadahiro M, Tabayashi K. Enhancement of heat shock protein expression after transient ischemia in the preconditioned spinal cord of rabbits. J Vasc Surg 1998; 27: 720–5.[Medline]

15 Svensson LG, Crawford ES, Hess KR, Coselli JS, Safi HJ. Experience with 1509 patients undergoing thoracoabdominal aortic operations. J Vasc Surg 1993; 17: 357–68; discussion 368–70.[Medline]

16 Ueno T, Itoh T, Hirahara K, Sakai M, Naitoh K. Protection against spinal cord ischaemia: one-shot infusion of hypothermic solution. Cardiovasc Surg 1994; 2: 374–8.[Medline]

17 Kakimoto M, Kawaguchi M, Sakamoto T, et al. Evaluation of rapid ischemic preconditioning in a rabbit model of spinal cord ischemia. Anesthesiology 2003; 99: 1112–7. Comment in: Anesthesiology 2004; 101: 261–2; author reply 262.[Medline]

18 Zvara DA, Colonna DM, Deal DD, Vernon JC, Gowda M, Lundell JC. Ischemic preconditioning reduces neurologic injury in a rat model of spinal cord ischemia. Ann Thorac Surg 1999; 68: 874–80. Comment in: Ann Thorac Surg 2000; 70: 2186.[Abstract/Free Full Text]

19 Sirin BH, Ortac R, Cerrahoglu M, et al. Ischaemic preconditioning reduces spinal cord injury in transient ischaemia. Acta Cardiol 2002; 57: 279–85.[Medline]

20 Iliodromitis EK, Kremastinos DT, Katritsis DG, Papadopoulos CC, Hearse DJ. Multiple cycles of preconditioning cause loss of protection in open-chest rabbits. J Mol Cell Cardiol 1997; 29: 915–20.[Medline]

21 Yang XM, Proctor JB, Cui L, Krieg T, Downey JM, Cohen MV. Multiple, brief coronary occlusions during early reperfusion protect rabbit hearts by targeting cell signaling pathways. J Am Coll Cardiol 2004; 44: 1103–10.[Abstract/Free Full Text]

22 Zvara DA, Bryant AJ, Deal DD, et al. Anesthetic preconditioning with sevoflurane does not protect the spinal cord after an ischemic-reperfusion injury in the rat. Anesth Analg 2006; 102: 1341–7.[Abstract/Free Full Text]

23 Contreras IS, Moreira LF, Ballester G, et al. Immediate ischemic preconditioning based on somatosensory evoked potentials seems to prevent spinal cord injury following descending thoracic aorta cross-clamping. Eur J Cardiothorac Surg 2005; 28: 274–9. Comment in: Eur J Cardiothorac Surg 2006; 29: 126; author reply 127.[Abstract/Free Full Text]

24 Dong H, Xiong L, Zhu Z, Chen S, Hou L, Sakabe T. Preconditioning with hyperbaric oxygen and hyperoxia induces tolerance against spinal cord ischemia in rabbits. Anesthesiology 2002; 96: 907–12.[Medline]

25 Duilio C, Ambrosio G, Kuppusamy P, DiPaula A, Becker LC, Zweier JL. Neutrophils are primary source of O2 radicals during reperfusion after prolonged myocardial ischemia. Am J Physiol Heart Circ Physiol 2001; 280: H2649–57.[Abstract/Free Full Text]

This Article Abstract Résumé de cet Article Full Text (PDF) Submit a scholarly reply Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Huang, H. Articles by Liu, J. Search for Related Content PubMed PubMed Citation Articles by Huang, H. Articles by Liu, J.